Analyzing sequential categorical data on dyadic interaction: A comment on Gottman.

Recent theoretical developments emphasize that social interactions are dynamic and reciprocal, and this has led to widespread use of time-series data on behavior in 2-person systems. In principle, such data allow one to separate the influences of 2 actors on each other. Statistical methods currently being used, however, are deficient in several respects. The present article shows that (1) a statistic proposed by G. P. Sackett (1979) and later "proved" by J. M. Gottman (see PA, Vol 62:5153; and 1980) is incorrect, and (2) the failure to control for autodependence can produce misleading results. New procedures are presented that are based on both traditional and more recently developed methods for the analysis of contingency tables. Using these procedures, methods are shown for testing dependency, measuring dependency, testing for differences in dependency across subgroups, and testing for dominance in reciprocal behavior sequences. (33 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)